A Test. How much Voltage (power) do your speakers need?

I measured the test tone at:

  • 2 volts or less

    Votes: 334 40.6%
  • Between 2-5 volts

    Votes: 252 30.7%
  • Between 5-10 volts

    Votes: 106 12.9%
  • Between 10-20 volts

    Votes: 55 6.7%
  • Over 20 volts.

    Votes: 75 9.1%

  • Total voters
    822
I happen to have the ASHRAE handbook of fundamentals on hand. It describes 70 db as "loud," 90 db as "Very loud," 110 db as "Threshold of discomfort" and 120 db as "Threshold of pain."

Most of my club PA sound experience has been in the threshold of discomfort to pain category. Although, after a while, you tend to accept it as normal (especially when weighed against other positive environmental variables :)).

Distortion or not, these levels are extremely loud, and I cannot see the point of striving to hit these levels in the home environment. In other words, it doesn't matter if your amp can swing that much voltage or not, it's almost never going to happen at home. Which brings us nicely to the point this thread is trying to make. You don't need megawatt amps for home use.

The “Threshold of Pain” is completely frequency dependent.

I find levels above about 100 dB at frequencies above 400 Hz or so to be uncomfortable, and seldom mix above 90 dBA but 120 dB SPL at frequencies below around 100 Hz does not cause me any pain.

Assuming reasonable stage volume, one can easily mix concerts with LF 10-20 dB over HF levels, providing a tactile feel without ear damaging HF levels. At home, where there is no “stage volume” to compete with, one can EQ to taste.

For those individuals with low sensitivity 8 ohm speakers that like the sound and feel of music at club or concert level bass at home, (evidently not many of those responding to this thread) amps capable of around 90 volts, around 1 kilowatt, would not be out of line.

An old (departed) friend of mine used to make a good living selling systems capable of high SPL, primarily to pro basketball players who liked to party hearty in the rather big houses they could afford.

If the test tones provided in the OP were 30 Hz and 60 Hz, the test results would be far different than they are now.
Of course, most of the testers would find their speakers flapping at 30 Hz before they would sound "loud" at all ;).
 
Administrator
Joined 2004
Paid Member
If the test tones provided in the OP were 30 Hz and 60 Hz, the test results would be far different than they are now.
No, they would not - unless you use heavy EQ or an active system. Think about it a minute. If you understand the test, you'll see that they would be pretty much the same, or identical.

BTW, I did provide a 30-60Hz sweep a few pages back.
 

ra7

Member
Joined 2009
Paid Member
If your objective was to hit 110 db peaks, and you had 80db/W speakers, that's 30db of headroom, equal to 1000 W.

At 90db/W speakers, or 20db headroom, you need a 100 W amp. But as Art and others have correctly pointed out, by the time you hit 110db, your speakers (and ears) are in much more trouble than your amp.

This works out very well for single-ended 2A3 backloaded horn aficionados. Assuming 95db/W speakers and a 100db target, which is still quite loud, you only need a 3W amp.
 
No, they would not - unless you use heavy EQ or an active system. Think about it a minute. If you understand the test, you'll see that they would be pretty much the same, or identical.

BTW, I did provide a 30-60Hz sweep a few pages back.
exactly, the "party" level was determined with music, not tones. using tones is not only stressful for the speakers and listener ears but would simply give a totally meaningless voltage.
 
The “Threshold of Pain” is completely frequency dependent.

We can say that. There is also a temporary protection mechanism, "stapedius reflex", that freezes the ossicles transmission chain of the middle ear. The resulting attenuation is about -20 dB but is not linear, mostly efficient for the 100/2000 Hz band. All mammals and birds have the device.

For the security of everybody's ears, never forget that this system shows a little delay before it's activation, and more, it is very short living:

@ 121dB 7 s.
@ 109 dB, 1 minute 52 s.
@ 100 dB, 15 minutes average...note that "only" 100 dB triggers the reflex.
 
PS it clips at 22V into 4 ohm. and that is 22V RMS. so it's more that 100 W per channel actually.

Which is a completely different thing to what was mentioned before. You mentioned 37 volt rails with output clipping at 22 volts, yet you were comparing an RMS figure against a 'peak' figure.

The rail voltages indicate the peak power that an amplifier is capable of producing. IE the 40 volt rails in a 100 watt amplifier will give 40^2/8 = 200 watts on a given transient, but will, as mentioned before, only give 100 watts with an rms voltage figure, ie 28Vrms. IE 10log(200/100) = 3dB more on a peak then with a steady state.

In your case with 37 volt rails, we are talking about a class D amplifier which probably uses a BTL output which will double the output voltage swing.

Given that you say your amplifier clips at 22Vrms then this gives us a peak figure of 31v, which is about right depending on the amplifier and any rail sag that happens to accompany the loading into 4 ohms.
 
Hi Art
One of the things I have been talking about here is that the dynamic range of music can vary from type to type by such a large amount that this test does not take this into account.

I was wondering if you have ever looked at the amplifier output voltage while playing music and watching the level indicators?

Crown used to have the Idep or whatever it was (like the psa-2 etc) that made the light change if there was even an instantaneous difference between the input and output wave shape.
It was subjectively much too sensitive and was ignored in favor of the nice red “getting your monies worth” light that came on later.

About 10 or 12 years ago, QSC had an A/B/X test rig going around doing amplifier demo’s and they came by the shop when worked in Glenview. We collected a number of pro amplifiers to use as comparisons and I brought my Threshold stasis II in from home that I used as my reference. The only thing special it had was it could drive my electrostatic speakers or normal speakers and didn’t seem to care either way.

Using my music choices and after sitting for a while, the amplifiers seemed to fall into two groups and one exception which was my threshold.
It sounded exactly like the others in one group except that with some things, it sounded less dynamic.

On the fast looking VU indicators, the peaks were -10 to -15, never any clip lights. Being puzzled, I got a scope and looked at the outputs. At around the time it sounded different (and not that loud), the output Voltage was reaching the rails and flat topping on very short single transients. This was NOT audible AT ALL unless one was comparing to an unclipped signal and then one heard it as being less dynamic.

At that point, I switched to a larger pro-sound amp (then a pl-236) and have not gone back. I was taken back by that experience and since then often enough one can find not only amp limit issues but gain structure problems.
I think there is some confusion too on the title of this since what is needed depends on many things and in spite of many attempts to differentiate loudness from dynamic range I have been unable at least for many. Feels a lot like the old days trying to explain extension is not the same as loud bass haha.

Lets say one does find it takes 2Volts RMS to produce a sine wave test signal like here as loud as you would care to have it. With a sine wave, all you need is an amplifier than supply that 2VRMS and no more.
So let’s say your amplifier can swing the suggested 4X or 8Volts RMS and all is well.

Now, one puts in AES pink noise at 2VRMS, to supply this signal, one needs an amplifier than can swing 4VRMS because this signal’s envelope has a 6dB peak to average ratio over the sine. Use pink noise from Smaart or any RF test gear used in audio and one finds about 10dB peak to average ratio in the envelope. This test signal requires an amplifier than can deliver 6.6VRMS in order to produce the 2VRMS signal.

As you play older recordings before the loudness wars or modern ones that aren’t compressed, the peak to average ratio gets larger than this and one either clips instantaneous peaks (which if short enough you can’t hear like normal clipping) or turn the average level down.
For an extreme example if one plays either the Harley recording or fireworks recording at the same volume setting as Velvet revolver, one will hardly be able to hear it, yet they both would have the same digital peak levels and if you turn the gain up, you will be raising the peak level accordingly and then clipping it off at 8V.

By the time speakers sound strained or clipped, one is massively nonlinear measurement wise and I am talking about faithful reproduction of the input signal.

My futile point has been that one must consider both the dynamic range of what one is producing AND the desired average loudness in order to estimate if one has enough power (because if you don’t, the peaks will be clipped and you may never know it).
The ONLY way to know in a given circumstance is to look with a scope.
Best,
Tom
 
Concert P.A.s are another matter, I'd love to see what voltages Tom is pushing for stadium sound. I can't tell you how much I use because all my P.A. speakers are powered (amp inside the speaker).
The B&C 21 inch 4 ohm speakers used in DSL's larger offerings are good for around 130 Volt peaks. They will take more, but prudent installers limit them to around that voltage.

More cabinets obviously require more amplifiers, but the voltage stays the same.

Your PA speakers should have specifications showing the peak power, when you see the clip/limit lights go on, you are probably there ;).

Art
 
Administrator
Joined 2004
Paid Member
One of the things I have been talking about here is that the dynamic range of music can vary from type to type by such a large amount that this test does not take this into account.
Yes Tom, it does. It has been stated and explained over and over again. This is your fundamental objection to the test, and it has been explained many times.
AndrewT seems to have the same objection and also has missed the point.

Your objection to the variety of dynamic range on recordings vs this test has been addressed again and again. No one denies the importance of dynamic range across many recordings, but the issue has been explained by me and others.

Why don't you try the test on your home system? We would be happy to see your results. Pick a nice dynamic recording and play it as loud as you ever do. Then measure the test tone and report it here. It's not a difficult test. :)
 
Administrator
Joined 2004
Paid Member
Can't you check from current draw....?
Maybe, but it would be tricky. I don't know how efficient the amps are, tho I might be able to guess by measuring the mains idle current vs current on a signal.

But it really doesn't matter because I have no control over it. The speakers (mostly Meyer Sound) come with amps and active crossover inside. All I can do is supply a line level signal at the back and let the electronics inside do its thing.
 
Hi Art
One of the things I have been talking about here is that the dynamic range of music can vary from type to type by such a large amount that this test does not take this into account.

I was wondering if you have ever looked at the amplifier output voltage while playing music and watching the level indicators?

By the time speakers sound strained or clipped, one is massively nonlinear measurement wise and I am talking about faithful reproduction of the input signal.

My futile point has been that one must consider both the dynamic range of what one is producing AND the desired average loudness in order to estimate if one has enough power (because if you don’t, the peaks will be clipped and you may never know it).
The ONLY way to know in a given circumstance is to look with a scope.
Best,
Tom
Your point has not been lost on me!

I don’t have an scope, so I rely on the amps lying red lights, my ears and a dB meter to measure actual music peak output.

Like you, I have found amplifier's meters don’t tell the whole story, I was surprised to see the old PSA 2 odep light coming on while it surpassed the SPL output of a modern amp rated about double it’s power.

The QSC PL-series has great peak power, the Crest CA series does too, while both Crest and QSC have offerings that just don’t cut the “swing test” for low frequencies.

Brian Oppegard, formerly of QSC, has really made a good sounding seven pound sub amp (the Torpedo SP-4000), if he can get a few changes implemented that will increase the rail voltage, it will make the high power B&C stand up and talk, as it stands the CA-9 in mono has a tad more swing.

I wish I could justify the expense of going to some of the other new lightweight offerings that have more peak potential.

Art
 
For an extreme example if one plays either the Harley recording or fireworks recording at the same volume setting as Velvet revolver, one will hardly be able to hear it, yet they both would have the same digital peak levels and if you turn the gain up, you will be raising the peak level accordingly and then clipping it off at 8V.

By the time speakers sound strained or clipped, one is massively nonlinear measurement wise and I am talking about faithful reproduction of the input signal.

My futile point has been that one must consider both the dynamic range of what one is producing AND the desired average loudness in order to estimate if one has enough power (because if you don’t, the peaks will be clipped and you may never know it).
The ONLY way to know in a given circumstance is to look with a scope.
Best,
Tom
I'm looking at it in a different way.
I measured the voltage where my amp starts clipping by analyzing the output waveform. that is with a full scale (0 dBfs) sine input. I remembered the volume knob position where that happens and therefore I know that as long as the volume knob is below that specific position, nothing will be clipped, no matter what I listen to, be it pink noise, sine waves, heavy metal or classical.
if I turn it past that position I know that sooner or later clipping will happen. simples.
 
I'm looking at it in a different way.
I measured the voltage where my amp starts clipping by analyzing the output waveform. that is with a full scale (0 dBfs) sine input. I remembered the volume knob position where that happens and therefore I know that as long as the volume knob is below that specific position, nothing will be clipped, no matter what I listen to, be it pink noise, sine waves, heavy metal or classical.
if I turn it past that position I know that sooner or later clipping will happen. simples.

Your kidding right. ........!